You answered your own question. Twitch is solid and well-built. It doesn't need an army of well-paid engineers to maintain it and isn't seeing a great ROI on new features. Some companies repurpose all, many or some engineers when the spreadsheet says to, others fire them wholesale.
Not many people realise this. Engineers build stuff. When your product is feature complete, you can get rid of most of them.
Big tech companies keeping tens of thousands of engineers on staff despite their core product not seeing substantial changes for years are just wasting money.
No growing company has a feature-complete product, as just load increases can make a product go from fine to inoperable. Sometimes it's not even more users, but more historical data: Something that used ot parse through 3 months of history is probably going to be unhappy with 5 years.
In big tech, having services without anyone having a semblance of maintenance responsibilities tends to lead to relatively quick failures. Even in boring enterprise companies, handing out maintenance to a 3rd party is still going to cost you, and is likely going to lower quality. I've seen way too many companies that ended up rewriting things after leaving a system mostly unmaintained for 5 years made it sop being fit to purpose.
If there's a lesson of software from the last decade is that undermaintained services become zombified software, and can eat your company's brains. That's why a place like google often would rather shut something down than claim it's finished.
No software product that has actual paying users can afford to stop development like that. It needs to constantly keep evolving with the environment, or someone else will fill their niche quickly or it will just atrophy. That may require fewer engineers though, so I agree with you partially.
With software, once you find PMF, it's not the building that's hard, but keeping the users paying and using your product. The entry barrier to building a copycat is very low. So you have to pile on resources to keep your software be the best available.
It's not a complete waste, every engineer you employ is one you don't have to compete with, which matters for big tech companies. Also, employed engineers can be mobilized to react to threats/emergencies. It's just a matter of time before they will be needed. If you are running a ghost ship, will you be able to scale up a response when needed? If the company is just being fleeced, then the answer is "who cares?" I guess.
Most tech products are not well engineered, they are riddled with tech debt as per typical advice. So when a product does succeed, it is usually limping along racing to add features and handle great load while being knee deep in (already) legacy code.
This is all fine and expected, you sacrifice engineering quality for development speed in the early days. Just expect to pay it back by keeping a large engineering team for many years to come, or see your entire tech explode in flames.
Those 'years' are finite however, Twitch has been mature for like a decade already. So finally the tech debt is all paid off, and given failure to develop major successful features, time to cut the staff.
When you're at scale, a .5% increase in performance or decrease in utilization pays for a team of engineers. Being able to have one less SaaS product bill as well. Maybe Twitch isn't at that scale, but some companies certainly are.
No company actually does this, they are just bloated because they have so much money and managers rank/value is determined by how many people they manage.
They explicitly did this in the 00's. Some circles of top companies over a decade ago were convicted for having anti-poaching agreements amongst each other.
May be a bit different now (as others said, free money is gone, so competition cropping up is much harder to do and much less a concern), but the tech boom of the 2000's very much worried about becoming the next MySpace to some startup's Facebook. That's why tech salaries became so high to begin with.
The anti-poaching thing was something completely different. That was preventing other companies from hiring people away from them because they were useful employees, not hiring people to prevent them from competing. Also, anti-poaching agreement kept salaries lower, it didn't inflate them.
It goes both ways. They can't collude woth hundreds of studios so they offer top salaries to out-compete the bulk of competition. Then for the remaining studios who can pay, they sign the agreements so they don't keep trying to one-up each other. Which would be bad for their business if they start trying to bid 300k for every worker (which may be what they deserve, but companies will always penny pinch).
>That was preventing other companies from hiring people away from them because they were useful employees, not hiring people to prevent them from competing.
Aren't these synonymous? You either entrap a necessary employee with contracts and/or absolute top compensation becsuse they are useful or even vital. No one's colluding to keep a janirot wage low (well, society is, but it's no one entity you can sue).
They have more competition than ever with youtube streaming actually somewhat catching on and tiktok streams becoming more popular. It seems weird to cut so many engineers when it’s probably most important now to innovate. Its not even just R&D that’s important, either. Their product is worse in some aspects than their competitors and a lot of the reason why they’re still dominant in their space is just the community of users on the website.
